Evaluating Fast Algorithms for Convolutional Neural Networks on FPGAs基于FPGAs的卷积神经网络快速算法评价.pdf

2017 IEEE 25th Annual International Symposium on Field-Programmable Custom Computing Machines Evaluating Fast Algorithms for Convolutional Neural Networks on FPGAs ∗ 1,3 † 1 1 2 ,3 Liqiang Lu , Yun Liang , Qingcheng Xiao , Shengen Yan 1Center for Energy-efficient Computing and Applications, Peking University, Beijing, China 2Department of Information Engineering, The Chinese University of Hong Kong. 3SenseTime Group Limited. Email: {liqianglu, ericlyun, walkershaw}@, yanshengen@ Abstract—In recent years, Convolutional Neural Networks A CNN typically involves multiple layers, where the (CNNs) have become widely adopted for computer vision output feature maps of one layer are the input feature maps tasks. FPGAs have been adequately explored as a promising of the following layer. Prior studies have shown that the hardware accelerator for CNNs due to its high performance, energy efficiency, and reconfigurability. However, prior FPGA computation of the state-of-the-art CNNs are dominated solutions based on the conventional convolutional algorithm by the convolutional layers [6, 7]. Using the conventional is often bounded by the computational capability of FPGAs convolution algorithm, each element in the output feature (e.g., the number of DSPs). In this paper, we demonstrate map is computed individually by using multiple multiply- that fast Winograd algorithm can dramatically reduce the accumulate operations. While the p